Strategy for predicting
transport-based durability properties of
concrete based on DEM approach
Kai Li, L.B.N. Le, Piet Stroeven, Martijn Stroeven Civil Engineering and Geosciences Faculty,
Delft University of Technology, the Netherlands
RILEM international workshop on performance-based specification and control of concrete durability, Zagreb, Croatia, 11 – 13 June, 2014
Outline
•
Transport-based durability issues of concrete
•
Discrete element method (DEM)
•
Pore network characteristics of hydrated paste
•
Simulation of hydrated structure on nano-scale
Transport-based durability issues of concrete
Durability problems of concrete Fluid, ion transport in concrete
Transport-based durability issues of concrete Three types of pores in cement-based materials:
•Gel pores few nanometers
•Capillary pores 1 nm – 10 µm
•Macro-pores (air voids) lager than 10 µm
Jing Hu, PhD thesis, Delft, 2004
Available techniques for pore exploration: • MIP (Mercury Intrusion Porosimetry) • IA (Images Analysis)
Transport-based durability issues of concrete
Macroscale Mesoscale Microscale Nanoscale
Discrete element method (DEM)
Newtonian system with particles (HADES package)
Linear direction: i i ij i j v F f m t
Angular direction: i i ij i j M m I t
Pore network characteristics of hydrated paste
HADES: dynamic DEM for packing simulation of fresh cement grains (Rosin-Rammler distribution).
Pore network characteristics of hydrated paste
XIPKM: extended integrated particle kinetics method for simulation of fresh multi-component cement system.
ro ro rout t = 0 t > 0 C3S C2S SiO2 CSH-in CSH-out CH Inert a in r b in r b a ro ro rout a a in r
(left) particle model for multi-component cement grain
(right) differently color-coded microstructure of hydrated paste
Pore network characteristics of hydrated paste
DRaMuTS: double random multiple tree structuring system for delineating the complete pore network system and for
topology assessment.
Different colors is for trees from different seeds
The main trunks in continuous pore channels Illustration of DRaMuTS
Pore network characteristics of hydrated paste
points in percolated pores points in isolated pores
points in main channels points in dead-end pores
Pore network characteristics of hydrated paste
SVM: star volume measurements in uniformly random point system inside the pore network
Pore size distribution in 1-30 m fully hydrated PC with w/c=0.3
C1: 6P C2: 4P+2R
P: periodic boundaries R: rigid boundaries
Pore network characteristics of hydrated paste
Validation
Hydration curves of PC (left) and porosity curves of RHA-blended PC samples for prolonged hydration (right). Note that 40/25 means wc=0.4/0.25; 10/20 % PC blending.
Simulation of hydrated structure on nano-scale However, “overestimation” of pore size distribution does exist……
Besides, for investigating transport-based durability properties of cementitious materials, surface roughness of hydration
product needs to be taken into account……
If we move from micro-level to nano-scale, probably solutions to our problems can be found……
Simulation of hydrated structure on nano-scale
HD C-S-H LD C-S-H
C3S
Assumed boundaries between HD and LD C-S-H
LD C-S-H HD
C-S-H
I.G. Richardson, Cement and Concrete Research, 34 (2004) 1733-1777
Simulation of hydrated structure on nano-scale
• C-S-H (calcium silicate hydrate), the main hydration product, important but complex gel.
• Granular nature and fractal dimension of C-S-H is suggested by Jennings’ colloidal model, meanwhile, fibrous morphology is revealed by experimental techniques.
• New mechanism should be employed based on existing Information to produce more realistic product.
2D illustration of hydrated structure on nano-scale a) b) c) d) a: stage 1 b: stage 2 c: stage 3 d: stage 4
red: cement grain blue: outer C-S-H white: pore space
Simulation of hydrated structure on nano-scale
Structural analysis – pore size distribution
S. Diamond, Cement and
Concrete Research, 30 (2000) 1517-1525
Conclusions
• Micro-level porosimetry approaches are useful for durability estimation purposes.
• The full range of cement particles should be involved for estimating transport properties.
• The nano-level approach could provide more realistic structure and morphology product, compared to the classical vector approach.
Conclusions
• A realistic shape factor should be added to the resulting expressions.
• A multi-scale techniques should be developed to bridge the gap between micro- and nano-level.